When a system failure occurs, information necessary to determine the cause must be recorded for service. This information includes data and control information in storage. In a single system, this storage may be in private address spaces or in common area. Units of work which may alter this storage are stopped while storage content is recorded. In particular, all address spaces, except those which are not needed, are set non-dispatchable. When the storage has been captured, the address spaces are reset and made dispatchable by the operating system.
By stopping the units of work that use the resources, system performance is degraded, since there is a great deal of idle time. In a multisystem environment, work units which alter shared resources execute across operating system images. Control information exists in each of the sharing systems and data exists on a commonly accessible data store. Gathering information to service a failure is a problem since work units which alter the data exists across operating systems, control information about data access exists across operating systems, and the data exists on a commonly accessible data store and is not captured on a failure.
Therefore, a need exists for a technique for serializing coupling facility data by preventing access to the data, not by stopping the work that may be using the data. In particular, a need exists for a technique for queuing requests to use the data until the serialization is released and the data is available. A further need exists for a method and system for accessing information in a data processing system such that requests to use data are queued until serialization is released and then the requests are re-driven. A further need exists for a technique for determining the amount of time that all users of a particular structure can tolerate not having access to that structure. A further need exists for a method and system for overriding the predefined amount of time, if it is determined such step is desirable. A further need exists for a technique for continuing a dump by another system in the sysplex when communication is lost between one or more operating systems and the coupling facility. Further, a need exists for a technique for capturing data when one or more operating systems coupled to a coupling facility lose communication with the coupling facility.